blob: 309561f5f5350a730dd16add9eed113e4fdb19a6 [file] [log] [blame]
* Copyright 2017 Google Inc.
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
#ifndef SkShaderBase_DEFINED
#define SkShaderBase_DEFINED
#include "include/core/SkColor.h"
#include "include/core/SkFlattenable.h"
#include "include/core/SkMatrix.h"
#include "include/core/SkPoint.h"
#include "include/core/SkRefCnt.h"
#include "include/core/SkScalar.h"
#include "include/core/SkShader.h"
#include "include/core/SkSurfaceProps.h"
#include "include/core/SkTypes.h"
#include "include/private/base/SkNoncopyable.h"
#include <cstddef>
#include <cstdint>
#include <optional>
#include <tuple>
class SkArenaAlloc;
class SkColorSpace;
class SkImage;
class SkRuntimeEffect;
class SkWriteBuffer;
enum SkColorType : int;
enum class SkTileMode;
struct SkDeserialProcs;
struct SkStageRec;
namespace SkShaders {
* This is used to accumulate matrices, starting with the CTM, when building up
* SkRasterPipeline or GrFragmentProcessor by walking the SkShader tree. It avoids
* adding a matrix multiply for each individual matrix. It also handles the reverse matrix
* concatenation order required by Android Framework, see b/256873449.
* This also tracks the dubious concept of a "total matrix", in the legacy Context/shadeSpan system.
* That includes all the matrices encountered during traversal to the current shader, including ones
* that have already been applied. The total matrix represents the transformation from the current
* shader's coordinate space to device space. It is dubious because it doesn't account for SkShaders
* that manipulate the coordinates passed to their children, which may not even be representable by
* a matrix.
* The total matrix is used for mipmap level selection and a filter downgrade optimizations in
* SkImageShader and sizing of the SkImage created by SkPictureShader. If we can remove usages
* of the "total matrix" and if Android Framework could be updated to not use backwards local
* matrix concatenation this could just be replaced by a simple SkMatrix or SkM44 passed down
* during traversal.
class MatrixRec {
MatrixRec() = default;
explicit MatrixRec(const SkMatrix& ctm);
* Returns a new MatrixRec that represents the existing total and pending matrix
* pre-concat'ed with m.
[[nodiscard]] MatrixRec concat(const SkMatrix& m) const;
* Appends a mul by the inverse of the pending local matrix to the pipeline. 'postInv' is an
* additional matrix to post-apply to the inverted pending matrix. If the pending matrix is
* not invertible the std::optional result won't have a value and the pipeline will be
* unmodified.
[[nodiscard]] std::optional<MatrixRec> apply(const SkStageRec& rec,
const SkMatrix& postInv = {}) const;
* FP matrices work differently than SkRasterPipeline. The starting coordinates provided to the
* root SkShader's FP are already in local space. So we never apply the inverse CTM. This
* returns the inverted pending local matrix with the provided postInv matrix applied after it.
* If the pending local matrix cannot be inverted, the boolean is false.
std::tuple<SkMatrix, bool> applyForFragmentProcessor(const SkMatrix& postInv) const;
* A parent FP may need to create a FP for its child by calling
* SkShaderBase::asFragmentProcessor() and then pass the result to the apply() above.
* This comes up when the parent needs to ensure pending matrices are applied before the
* child because the parent is going to manipulate the coordinates *after* any pending
* matrix and pass the resulting coords to the child. This function gets a MatrixRec that
* reflects the state after this MatrixRec has bee applied but it does not apply it!
* Example:
* auto childFP = fChild->asFragmentProcessor(args, mrec.applied());
* childFP = MakeAWrappingFPThatModifiesChildsCoords(std::move(childFP));
* auto [success, parentFP] = mrec.apply(std::move(childFP));
MatrixRec applied() const;
/** Call to indicate that the mapping from shader to device space is not known. */
void markTotalMatrixInvalid() { fTotalMatrixIsValid = false; }
/** Marks the CTM as already applied; can avoid re-seeding the shader unnecessarily. */
void markCTMApplied() { fCTMApplied = true; }
* Indicates whether the total matrix of a MatrixRec passed to a SkShader actually
* represents the full transform between that shader's coordinate space and device space.
bool totalMatrixIsValid() const { return fTotalMatrixIsValid; }
* Gets the total transform from the current shader's space to device space. This may or
* may not be valid. Shaders should avoid making decisions based on this matrix if
* totalMatrixIsValid() is false.
SkMatrix totalMatrix() const { return SkMatrix::Concat(fCTM, fTotalLocalMatrix); }
/** Gets the inverse of totalMatrix(), if invertible. */
[[nodiscard]] bool totalInverse(SkMatrix* out) const {
return this->totalMatrix().invert(out);
/** Is there a transform that has not yet been applied by a parent shader? */
bool hasPendingMatrix() const {
return (!fCTMApplied && !fCTM.isIdentity()) || !fPendingLocalMatrix.isIdentity();
/** When generating raster pipeline, have the device coordinates been seeded? */
bool rasterPipelineCoordsAreSeeded() const { return fCTMApplied; }
MatrixRec(const SkMatrix& ctm,
const SkMatrix& totalLocalMatrix,
const SkMatrix& pendingLocalMatrix,
bool totalIsValid,
bool ctmApplied)
: fCTM(ctm)
, fTotalLocalMatrix(totalLocalMatrix)
, fPendingLocalMatrix(pendingLocalMatrix)
, fTotalMatrixIsValid(totalIsValid)
, fCTMApplied(ctmApplied) {}
const SkMatrix fCTM;
// Concatenation of all local matrices, including those already applied.
const SkMatrix fTotalLocalMatrix;
// The accumulated local matrices from walking down the shader hierarchy that have NOT yet
// been incorporated into the SkRasterPipeline.
const SkMatrix fPendingLocalMatrix;
bool fTotalMatrixIsValid = true;
// Tracks whether the CTM has already been applied (and in raster pipeline whether the
// device coords have been seeded.)
bool fCTMApplied = false;
} // namespace SkShaders
#define SK_ALL_SHADERS(M) \
M(Blend) \
M(CTM) \
M(Color) \
M(Color4) \
M(ColorFilter) \
M(CoordClamp) \
M(Empty) \
M(GradientBase) \
M(Image) \
M(LocalMatrix) \
M(PerlinNoise) \
M(Picture) \
M(Runtime) \
M(Transform) \
M(TriColor) \
M(Conical) \
M(Linear) \
M(Radial) \
class SkShaderBase : public SkShader {
~SkShaderBase() override;
sk_sp<SkShader> makeInvertAlpha() const;
sk_sp<SkShader> makeWithCTM(const SkMatrix&) const; // owns its own ctm
* Returns true if the shader is guaranteed to produce only a single color.
* Subclasses can override this to allow loop-hoisting optimization.
virtual bool isConstant() const { return false; }
enum class ShaderType {
#define M(type) k##type,
#undef M
virtual ShaderType type() const = 0;
enum class GradientType {
#define M(type) k##type,
#undef M
* If the shader subclass can be represented as a gradient, asGradient
* returns the matching GradientType enum (or GradientType::kNone if it
* cannot). Also, if info is not null, asGradient populates info with
* the relevant (see below) parameters for the gradient. fColorCount
* is both an input and output parameter. On input, it indicates how
* many entries in fColors and fColorOffsets can be used, if they are
* non-NULL. After asGradient has run, fColorCount indicates how
* many color-offset pairs there are in the gradient. If there is
* insufficient space to store all of the color-offset pairs, fColors
* and fColorOffsets will not be altered. fColorOffsets specifies
* where on the range of 0 to 1 to transition to the given color.
* The meaning of fPoint and fRadius is dependent on the type of gradient.
* None:
* info is ignored.
* Color:
* fColorOffsets[0] is meaningless.
* Linear:
* fPoint[0] and fPoint[1] are the end-points of the gradient
* Radial:
* fPoint[0] and fRadius[0] are the center and radius
* Conical:
* fPoint[0] and fRadius[0] are the center and radius of the 1st circle
* fPoint[1] and fRadius[1] are the center and radius of the 2nd circle
* Sweep:
* fPoint[0] is the center of the sweep.
struct GradientInfo {
int fColorCount = 0; //!< In-out parameter, specifies passed size
// of fColors/fColorOffsets on input, and
// actual number of colors/offsets on
// output.
SkColor* fColors = nullptr; //!< The colors in the gradient.
SkScalar* fColorOffsets = nullptr; //!< The unit offset for color transitions.
SkPoint fPoint[2]; //!< Type specific, see above.
SkScalar fRadius[2]; //!< Type specific, see above.
SkTileMode fTileMode;
uint32_t fGradientFlags = 0; //!< see SkGradientShader::Flags
virtual GradientType asGradient(GradientInfo* info = nullptr,
SkMatrix* localMatrix = nullptr) const {
return GradientType::kNone;
enum Flags {
//!< set if all of the colors will be opaque
kOpaqueAlpha_Flag = 1 << 0,
* ContextRec acts as a parameter bundle for creating Contexts.
struct ContextRec {
ContextRec(SkAlpha paintAlpha,
const SkShaders::MatrixRec& matrixRec,
SkColorType dstColorType,
SkColorSpace* dstColorSpace,
const SkSurfaceProps& props)
: fMatrixRec(matrixRec)
, fDstColorType(dstColorType)
, fDstColorSpace(dstColorSpace)
, fProps(props)
, fPaintAlpha(paintAlpha) {}
static ContextRec Concat(const ContextRec& parentRec, const SkMatrix& localM) {
return {parentRec.fPaintAlpha,
const SkShaders::MatrixRec fMatrixRec;
SkColorType fDstColorType; // the color type of the dest surface
SkColorSpace* fDstColorSpace; // the color space of the dest surface (if any)
SkSurfaceProps fProps; // props of the dest surface
SkAlpha fPaintAlpha;
bool isLegacyCompatible(SkColorSpace* shadersColorSpace) const;
class Context : public ::SkNoncopyable {
Context(const SkShaderBase& shader, const ContextRec&);
virtual ~Context();
* Called sometimes before drawing with this shader. Return the type of
* alpha your shader will return. The default implementation returns 0.
* Your subclass should override if it can (even sometimes) report a
* non-zero value, since that will enable various blitters to perform
* faster.
virtual uint32_t getFlags() const { return 0; }
* Called for each span of the object being drawn. Your subclass should
* set the appropriate colors (with premultiplied alpha) that correspond
* to the specified device coordinates.
virtual void shadeSpan(int x, int y, SkPMColor[], int count) = 0;
// Reference to shader, so we don't have to dupe information.
const SkShaderBase& fShader;
uint8_t getPaintAlpha() const { return fPaintAlpha; }
const SkMatrix& getTotalInverse() const { return fTotalInverse; }
SkMatrix fTotalInverse;
uint8_t fPaintAlpha;
* Make a context using the memory provided by the arena.
* @return pointer to context or nullptr if can't be created
Context* makeContext(const ContextRec&, SkArenaAlloc*) const;
* If the shader can represent its "average" luminance in a single color, return true and
* if color is not NULL, return that color. If it cannot, return false and ignore the color
* parameter.
* Note: if this returns true, the returned color will always be opaque, as only the RGB
* components are used to compute luminance.
bool asLuminanceColor(SkColor4f*) const;
* If this returns false, then we draw nothing (do not fall back to shader context). This should
* only be called on a root-level effect. It assumes that the initial device coordinates have
* not yet been seeded.
[[nodiscard]] bool appendRootStages(const SkStageRec& rec, const SkMatrix& ctm) const;
* Adds stages to implement this shader. To ensure that the correct input coords are present
* in r,g MatrixRec::apply() must be called (unless the shader doesn't require it's input
* coords). The default impl creates shadercontext and calls that (not very efficient).
virtual bool appendStages(const SkStageRec&, const SkShaders::MatrixRec&) const = 0;
virtual SkImage* onIsAImage(SkMatrix*, SkTileMode[2]) const {
return nullptr;
virtual SkRuntimeEffect* asRuntimeEffect() const { return nullptr; }
static Type GetFlattenableType() { return kSkShader_Type; }
Type getFlattenableType() const override { return GetFlattenableType(); }
static sk_sp<SkShaderBase> Deserialize(const void* data, size_t size,
const SkDeserialProcs* procs = nullptr) {
return sk_sp<SkShaderBase>(static_cast<SkShaderBase*>(
SkFlattenable::Deserialize(GetFlattenableType(), data, size, procs).release()));
static void RegisterFlattenables();
* If this shader can be represented by another shader + a localMatrix, return that shader and
* the localMatrix. If not, return nullptr and ignore the localMatrix parameter.
virtual sk_sp<SkShader> makeAsALocalMatrixShader(SkMatrix* localMatrix) const;
static SkMatrix ConcatLocalMatrices(const SkMatrix& parentLM, const SkMatrix& childLM) {
#if defined(SK_BUILD_FOR_ANDROID_FRAMEWORK) // b/256873449
return SkMatrix::Concat(childLM, parentLM);
return SkMatrix::Concat(parentLM, childLM);
void flatten(SkWriteBuffer&) const override;
* Specialize creating a SkShader context using the supplied allocator.
* @return pointer to context owned by the arena allocator.
virtual Context* onMakeContext(const ContextRec&, SkArenaAlloc*) const {
return nullptr;
virtual bool onAsLuminanceColor(SkColor4f*) const {
return false;
friend class SkShaders::MatrixRec;
inline SkShaderBase* as_SB(SkShader* shader) {
return static_cast<SkShaderBase*>(shader);
inline const SkShaderBase* as_SB(const SkShader* shader) {
return static_cast<const SkShaderBase*>(shader);
inline const SkShaderBase* as_SB(const sk_sp<SkShader>& shader) {
return static_cast<SkShaderBase*>(shader.get());
void SkRegisterBlendShaderFlattenable();
void SkRegisterColor4ShaderFlattenable();
void SkRegisterColorShaderFlattenable();
void SkRegisterCoordClampShaderFlattenable();
void SkRegisterEmptyShaderFlattenable();
void SkRegisterPerlinNoiseShaderFlattenable();
void SkRegisterWorkingColorSpaceShaderFlattenable();
#endif // SkShaderBase_DEFINED